MOS Jasmin Lau의 KPMG 싱가포르 Trusted AI Centre of Excellence 출범 연설

Ms Lee Sze Yeng, Managing Partner, KPMG

Mr Lyon Poh, Partner and AI CoE Lead, KPMG

Industry partners and colleagues,

Thank you for having me today.

Congratulations to KPMG on your 85th year in Singapore.

Now, 85 years is not just a milestone. It is proof that trust, which is earned by humans consistently over time, is a durable and competitive advantage. Today's launch of the Trusted AI Centre of Excellence (CoE) carries that same spirit into a new era.

I want to acknowledge the people in this room. The KPMG partners, the staff, the clients who have built this firm over many decades. 85 years of any institution is really 85 years of individual decisions to do the right thing even when sometimes it was easier not to. That is what credibility is made of. And it is also exactly the disposition Singapore needs as we navigate AI.

We are in the middle of something big, and everybody knows it. The workers, the citizens, and every company leader understands it.

If you ask any worker today - in a bank, in a clinic, on a manufacturing factory floor, many will tell you they are uncertain about what AI means for them.

If you ask the Minister of State Jasmin Lau, I would also tell you that I am not sure what AI would mean for myself as well as for Singapore. It is not a space where any of us can predict what is going to happen next.

And all of us are not alone. Around the world, we are seeing a growing unease about AI. We follow social media to see what is going on, we watch as across university graduation ceremonies, leaders are getting booed when they talk about AI. We are acutely aware of the anxiety that many countries and populations have about the technology.

Some of the anxiety stems from real harms, but a lot of it is more fundamental: people are worried about being left behind. They see the technology advancing fast, they hear about enormous gains in productivity and profit, and then they wonder - is that also for me?

Now, Singapore cannot afford to ignore this. And we will not. Many of us have spoken in Parliament about this issue, and we fully intend for the gains of AI to be shared broadly across the population.

Earlier this year, the Economic Strategy Review (ESR) committee set out our AI ambitions. We identified AI Missions across key sectors. We arecommitted to broad-based AI adoption — not just among leading firms, but also across our SMEs and our workforce.

The goal that we have is an AI-empowered economy where many companies and their workers all benefit. Not just the few who decide to move first.

But the ESR Committee on Technology and Innovation - which I co-chair with SPS Goh Hanyan – also made clear that trust is a core enabler. Without trust, adoption stalls. Not because the technology is not ready, but because our people are not.

Singapore's AI strategy will therefore only succeed if our people - workers and citizens - feel that they are part of it, not just subject to it.

For employees, we need to show them - through reskilling programmes, through workplace governance standards, through honest conversations about where changes will happen - that this transition is being managed with them in mind. Companies have a very big responsibility to play here.

For our citizens, it also means that AI systems have affected their lives - in public services, in healthcare, in financial decisions – all of these must be fair, explainable, and accountable. When an AI system influences whether you get a loan, a job shortlist, or a medical referral, that person has a right to understand why, and a right to challenge it if anything goes wrong.

This is not just about ethics. It is about the basic compact between a government, its institutions, and its people. I hope that the next time Ms Lee (Sze Yeng) asks a question about who is that someone you trust to have your back, more of you will say that you trust that the government has your back, even as we roll out AI systems across the public service.

This brings me to something many of you are already grappling with in this room. If governance requires assurance, and assurance requires audit, then how do we audit AI?

We know that traditional audit works by examining records, sampling transactions, testing controls. These are all proven methods. But AI systems are fundamentally different. They are complex, often opaque to many of us, and they evolve continuously. A model performing well six months ago may behave differently today because the data has shifted, the operating environment has changed, or the model has been retrained.

If you don’t own and design your own model, you may not even know that the model has changed. And so, if your audit cycle runs in months but your AI system updates in weeks or days, then your assurance is always behind.

To audit AI credibly and at scale, the audit function itself must be AI-powered, with continuous monitoring and proactive risk management, not periodic reviews or post-hoc investigations after something goes wrong. This matters especially when AI becomes more autonomous, making decisions with less human oversight.

Closing that gap requires firms that understand both the importance of technology and the discipline of assurance. That is a rare combination, but that is also what KPMG has done over the years.

That is what KPMG's Trusted AI Centre of Excellence is precisely designed to do. KPMG is applying AI to its own audit, tax, and advisory functions using itself as "client zero", the proving ground for what AI-powered assurance looks like. They are doing the hard work themselves, finding where the approach works and where it needs refinement. From that foundation, the CoE can then work with companies to deploy AI responsibly – creating scalable AI solutions with governance and trust built in from the start.

For Singapore as a country, we have been quite deliberate about building AI governance capabilities – again, not as an afterthought to our AI strategy, but right from the start, as a core part of it.

Companies operating across borders face a growing patchwork of standards. Singapore is actively working to shape and harmonise these standards - driving ASEAN-wide adoption of tools like AI Verify, and launching a programme to accredit third-party AI testers, so companies know who they can trust. The goal is to make Singapore and Singapore-based companies the benchmarks for trusted AI in the region, the place where standards are set, tested, and exported.

As more companies deploy AI across borders, the demand for credible, independent assurance will only increase. This is a space where Singapore can lead, and I am glad that this is also the space that KPMG will be playing in.

So, let me end where I began, with the workers and citizens who are watching how this unfolds.

They are not asking us to slow down. Most of them are asking to be included. They want to know that the gains from AI will be shared, that the risks will be managed, and that when something goes wrong, someone will be accountable.

Meeting that expectation is a shared responsibility - for government, and also for industry leaders like KPMG and others in this room.

Singapore is building the conditions for this: the governance frameworks, the regulatory clarity, the talent pipeline. KPMG's decision to anchor its Trusted AI Centre of Excellence here tells me we are on the right track.

But frameworks alone are not enough. The real test is whether the people of Singapore - workers retraining for new roles, citizens interacting with AI-driven services – the real test is for them to feel that the AI transformation is happening with them, not to them. We also have to make sure that our SME owners feel comfortable navigating the new and unfamiliar tools, and the more effort all of us put into building this ecosystem, the more our SMEs and our population working in these SMEs will feel comfortable with the transition that is happening.

That is the standard we should hold all of ourselves to. And in a room like this one, with the expertise and the accountability that KPMG and its partners bring, I believe that we can meet it.

Thank you very much.